Réunion island in the Indian Ocean is commonly considered as the recent and active expression of the hotspot that formed the Deccan traps, although both the hypothesis of recent small hotspots for both Reunion and Mauritius, or of relation with the plate heterogeneity have been proposed. Structural studies by seismic methods, from the scale of the upper cone of the active Fournaise volcano to that of the crust 100 km around, have been carried out. At this scale significant departures appear from the Hawaiian case to which it is traditionally compared, with the seismic signature of active volcanism showing differences too. Refraction-reflection seismics do not see a geometry of the top of the underlying plate towards the island, expected in plate flexure modelling by analogy with other hotspot island. Where it is sampled, doming is suggested instead. There appears to be less magmatic products than if there was a large amount buried in a flexural depression. The velocity structure resolved for the volcanic island, apart from high velocity cores under the volcanoes leads to smaller overall density than usually considered in flexure modelling. The same appears to hold for the material of the cone of about 120 km radius rising above the regional sea-bottom level to the 30 km radius island, from coincident reflection and refraction seismics on several lines radial to the southeastern half of the island. At the crust-mantle level, there is evidence from reflection-refraction line extending 150 km either side of the island for a layer of velocity intermediate between normal crust and mantle values. Two radial reflection line to the SSW, close to each other detect a differences in depth of the oceanic basement. This may coincide with a fracture zone suggested from the reconstruction of the sea-floor spreading history from the magnetic anomaly pattern. The latter has been interpreted previously to indicate that the western part of Réunion developed atop a Paleogene fossil accretionary center. This has also been suggested for Mauritius island, the two fossil accretionary center having been active on different sides of a triple junction, and carried away from each other along the later fracture zone in-between. The available seismic data sample only from the top of the island to the top of the mantle. At this scale they evidence a departure of Réunion from an idealized oceanic hotspot volcanic islands model, as well as the relation of its location with a structural heterogeneity of the underlying lithosphere that appears inherited from its complex origin. These partly unexpected results suggest that the case deserves further sampling at the broader regional scale including Mauritius, and deeper into the mantle.